Case Study: Knowde achieves modern semantic search and improved product discovery for its chemicals B2B marketplace with Grid Dynamics

Semantic search   for the chemicals B2B marketplace

Knowde is a digital B2B marketplace for chemicals, ingredients and polymers that connects R&D and procurement teams with thousands of suppliers and products. Knowde faced a search challenge: its original monolithic, keyword-based search couldn’t leverage a rich, domain-specific taxonomy and struggled with chemical tokenization, entity recognition and relevance across very different user intents. Grid Dynamics was engaged to modernize Knowde’s search capabilities, moving from baseline keyword matching to a semantic, concept-aware search packaged business capability (PBC).

Grid Dynamics implemented a semantic search PBC—including semantic query parsing, domain-specific query expansion using knowledge graphs, intent classification, custom grammar-based tokenization and Elasticsearch extensions—and launched the solution in six months. The new Grid Dynamics search cut average results returned from ~27,000 to ~700, improved first-page relevance from 51% to 92%, increased QCTR from 15.8% to 20.5% and CTR from 20.37% to 26.48% (≈30% uplift for both), reduced page changes by 60% and search impressions by 40%, and drove a 60% increase in category selection with 30% lower filter usage.

Open case study document...